
***************************************************************************
*********************** REPLICATION DOFILE ********************************
***** Local Context and Mobilization in Poor Communities ******************
***************** Prisca Jöst and Ellen Lust ******************************
***************************************************************************

clear all

set maxvar 32000

use "REPLICATION_DATA_FINAL.dta", clear

*********************************************
**** PREPARATION *******************
*********************************************

*How many SQKM have n>20?
bysort sqkm: gen sqkm_freq= _N

*Which are those?
gen less20sqkm= 0 if sqkm_freq!=.
replace less20sqkm=1 if sqkm_freq<20

encode sqkm, gen(n_sqkm)

*Now exclude those from the data
drop if less20sqkm==1


**** Countries: Generate country variable ****

*** Zambia ****

gen Zambia_c = .
replace Zambia_c = 1 if LGPI_region=="Zambia" | LGPI_region=="Lusaka"
replace Zambia_c = 0 if LGPI_region=="Malawi" | LGPI_region=="Lilongwe" | LGPI_region=="Nairobi"

*** Malawi ****

gen Malawi_c = .
replace Malawi_c = 1 if LGPI_region=="Malawi" | LGPI_region=="Lilongwe"
replace Malawi_c = 0 if LGPI_region=="Zambia" | LGPI_region=="Lusaka" | LGPI_region=="Nairobi"

** Kenya ***
gen Kenya = .
replace Kenya = 1 if LGPI_region=="Nairobi"
replace Kenya = 0 if LGPI_region=="Zambia" | LGPI_region=="Lusaka" | LGPI_region=="Malawi" | LGPI_region=="Lilongwe"

******************************************************
***** DEPENDENT VARIABLES *******
******************************************************

* selfcompliance is coded with 4 as very likely in the data set *

* How likely is your leader to sanction you if you do not comply *
gen leadersanction = 5-caul_q10 if caul_q10<5

* How likely is others in the village to sanction you if you do not comply *
gen villagesanction = 5-caul_q11 if caul_q11<5

* If you knew that enough other people are participating in this activity to make it successful, would this make you more likely to comply? *
gen bandwagon = 4-caul_q12 if caul_q12<4

* How much do you think it is right and proper for your leader to ask *
gen legitimacy = 5-caul_q9 if caul_q9<5

* In general, how fair is your {leader/caul_q1} when dealing with you and other members of your village? *
gen fairness = 5-caul_q13 if caul_q13<5

* Do you think the community is better off if people listen to and follow what your leader*
gen benefit = .
replace benefit = 1 if caul_q14==2
replace benefit = 0 if caul_q14==1

* In general, do you think your {leader/caul_q1} is more interested in helping himself/herself and close friends, or in helping the village as a whole?*
gen leaderinterest = .
replace leaderinterest= 1 if caul_q15==1
replace leaderinterest = 0 if caul_q15==2

* Think about how many people in your village/neighborhood know your {leader/caul_q1} by name.  Would you say that it is almost everyone, some people, a few people, or hardly anyone?*
gen knowing = 5-caul_q16 if caul_q16<5

*How difficult is it for you to access your {leader/caul_q1}?*
gen access = 5-caul_q17 if caul_q17<5

***********************************************************
****** INDIVIDUAL MEASURES *****
***********************************************************

*Social ties: Know some or most people in the neighborhood/village 

gen knowpeople= 0 if vnin_q3!=. & vnin_q3!=5
replace knowpeople= 1 if (vnin_q3==4|vnin_q3==3)

*Insufficient income 

gen insuffinc= 0 if wealth1!=. & wealth1!=5
replace insuffinc=1 if (wealth1==3 | wealth1==4)


******* SQKM MEASURES (CONTEXTUAL VARIABLES) ***********

*Here the measures on the sqkm level will be calculated: Share of Respondents who know some or most people 

bysort sqkm: egen knowpeopleshare= mean(knowpeople)
histogram knowpeopleshare, frequency

*Density: share by sqkm using 80, 60, 70 and 85 percent threshold

gen knowpeoplesqkm_binary= 0 if knowpeopleshare!=.
replace knowpeoplesqkm_binary= 1 if knowpeopleshare>0.79

gen knowpeoplesqkm_binary2= 0 if knowpeopleshare!=.
replace knowpeoplesqkm_binary2= 1 if knowpeopleshare>0.59

gen knowpeoplesqkm_binary3= 0 if knowpeopleshare!=.
replace knowpeoplesqkm_binary3= 1 if knowpeopleshare>0.69

gen knowpeoplesqkm_binary4= 0 if knowpeopleshare!=.
replace knowpeoplesqkm_binary4= 1 if knowpeopleshare>0.84


* Insufficient income by SQKM

bysort sqkm: egen sqkinsuffinc= mean(insuffinc)

histogram sqkinsuffinc, frequency

gen sqkminsuffinc_binary= 0 if sqkinsuffinc!=.
replace sqkminsuffinc_binary=1 if sqkinsuffinc>0.79

**************************************************
*************** INDIVIDUAL CONTROLS ***************
**************************************************

* CtrlAge

gen female = .
replace female = 1 if gender == 2
replace female = 0 if gender == 1
label define female 1 "female" 0 "male"
label values female female

** education **
gen educ2 = .
replace educ2 = 0 if education == 1
replace educ2 = 1 if education > 1
label define educ2 0 "no schooling" 1 "at least primary schooling"
label values educ2 educ2

** lived **
gen lived2 = .
replace lived2 = 1 if lived == 1
replace lived2 = 2 if lived == 2 | lived == 3 | lived == 4
replace lived2 = 3 if lived == 5
label define lived2 1 "less than 1 year" 2 "more than 1 year" 3 "all my life"

*** Asked by village head or neighbor *****

gen neighbor_VH = 0 if caul_q1<=5
replace neighbor_VH = 1 if caul_q1==1 | caul_q1==5


** Asked by village head ****

gen VH = 0 if caul_q1<=5
replace VH = 1 if caul_q1==1

*** Asked by neighbor ***
gen neighbor = 0 if caul_q1<=5
replace neighbor = 1 if caul_q1==5

*** Asked by MP ***
gen MP = 0 if caul_q1<=5
replace MP = 1 if caul_q1==4

*** Asked by TA ***
gen TA = 0 if caul_q1<=5
replace TA = 1 if caul_q1==2

*** Asked by councilor ***
gen councilor = 0 if caul_q1<=5
replace councilor = 1 if caul_q1==3

** activities ***

gen voting = 0 if caul_q2<=3
replace voting = 1 if caul_q2==1

gen burialfund = 0 if caul_q2<=3
replace burialfund = 1 if caul_q2==2

gen schoolfund = 0 if caul_q2<=3
replace schoolfund = 1 if caul_q2==3


** encode regions variable ***

encode LGPI_region, gen(n_LGPI_region)



********************************************************************************
*Drop if insuficient income (poor) == 0 to only look at the poor in the sample*
********************************************************************************

drop if insuffinc == 0


********************************************************************************
*********************** ANALYSIS **************************************
********************************************************************************

* Empty Model (the following results are reported in Table E4-6 in the Appendix) *

*mixed selfcompliance || sqkm:, covariance(independent)
*estimates store Nullmodel

*estat icc

* Random Intercept Model *

*mixed selfcompliance neighbor_VH || sqkm:, covariance(independent)
*estimates store intercept

* Random Slope Model *

*mixed selfcompliance neighbor_VH || sqkm: neighbor_VH, covariance(independent)
*estimates store randomslope

*lrtest intercept randomslope


**********************************************************************
******* Table 1 in Main Text AND FULL MODELS IN APPENDIX E. ******
**********************************************************************

** Add level 1 variables 

mixed selfcompliance i.neighbor_VH i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

** Add level 1 and level 2 variables: not reported in main text **

mixed selfcompliance i.neighbor_VH i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region c.knowpeopleshare || sqkm: neighbor_VH, covariance(independent)

* Cross-level interaction model **
mixed selfcompliance c.knowpeopleshare##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

margins neighbor_VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))


**********************************************************************
******* Table 2 in Main Text AND FULL MODELS IN APPENDIX E. ******
**********************************************************************

** the results of the empty models, random intercept and random slopes models are not reported in the main text **

******** Empty Models ************

*mixed villagesanction || sqkm:, covariance(independent)
*estat icc

*mixed bandwagon || sqkm:, covariance(independent)
*estat icc

*mixed leadersanction || sqkm:, covariance(independent)
*estat icc

******* Random Intercept Models ********

*mixed villagesanction neighbor_VH || sqkm:, covariance(independent)
*estimates store intercept3

*mixed bandwagon neighbor_VH || sqkm:, covariance(independent)
*estimates store intercept4

*mixed leadersanction neighbor_VH || sqkm:, covariance(independent)
*estimates store intercept2

****** Random Slope Models **********

*mixed villagesanction neighbor_VH || sqkm: neighbor_VH, covariance(independent)
*estimates store randomslope3

*lrtest intercept3 randomslope3

*mixed bandwagon neighbor_VH || sqkm: neighbor_VH, covariance(independent)
*estimates store randomslope4

*lrtest intercept4 randomslope4

*mixed leadersanction neighbor_VH || sqkm: neighbor_VH, covariance(independent)
*estimates store randomslope2

*lrtest intercept2 randomslope2

** Random Slope with level 1 and level 2 variables **

*mixed villagesanction i.neighbor_VH i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region c.knowpeopleshare || sqkm: neighbor_VH, covariance(independent)

*mixed bandwagon i.neighbor_VH i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region c.knowpeopleshare || sqkm: neighbor_VH, covariance(independent)

*mixed leadersanction i.neighbor_VH i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region c.knowpeopleshare || sqkm: neighbor_VH, covariance(independent)


******* Cross level interaction ****************************************
*** reported in Table 2 and Figure 2 in Main text and Appendix E ***

mixed villagesanction  c.knowpeopleshare##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

margins neighbor_VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))

mixed bandwagon c.knowpeopleshare##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

margins neighbor_VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))

mixed leadersanction c.knowpeopleshare##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

margins neighbor_VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))

****************************************************************************************
********* Subgroup analysis by type of activity: Main text Table 3 and Figure 3 and Appendix Tables C10-12 and Figures C1-3 *****

*** voting ***
*drop if PGtreat == 1 | burialtreat == 1


*** burial fund ****
*drop if PGtreat == 1 | votetreat == 1


*** educational fund ****
*drop if votetreat == 1 | burialtreat == 1

mixed selfcompliance i. neighbor_VH i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

*mixed selfcompliance i.neighbor_VH i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region c.knowpeopleshare || sqkm: neighbor_VH, covariance(independent)

mixed selfcompliance c.knowpeopleshare##i.neighbor_VH i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

margins neighbor_VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))

******* MECHANISMS ************

mixed leadersanction i. neighbor_VH i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

*mixed selfcompliance i.neighbor_VH i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region c.knowpeopleshare || sqkm: neighbor_VH, covariance(independent)

mixed leadersanction c.knowpeopleshare##i.neighbor_VH i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

margins neighbor_VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))


mixed bandwagon i. neighbor_VH i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

*mixed selfcompliance i.neighbor_VH i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region c.knowpeopleshare || sqkm: neighbor_VH, covariance(independent)

mixed bandwagon c.knowpeopleshare##i.neighbor_VH i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

margins neighbor_VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))


mixed villagesanction i. neighbor_VH i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

*mixed selfcompliance i.neighbor_VH i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region c.knowpeopleshare || sqkm: neighbor_VH, covariance(independent)

mixed villagesanction c.knowpeopleshare##i.neighbor_VH i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

margins neighbor_VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))


****************************************************************************************
****************************************************************************************
*** Anlaysis by village head/ neighbor: Table 4 and Figure 4 in Main Text and Appendix Tables C13-14 and Figures 4-5 *****
****************************************************************************************

*** Village head only: drop neighbor==1 ***
*drop if neighbor==1

mixed selfcompliance i.VH i.female i.educ2 ib1.lived2 i.CtrlAge i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: VH, covariance(independent)

*mixed selfcompliance i.VH i.female i.educ2 ib1.lived2 i.CtrlAge i.highPopulation i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region c.knowpeopleshare || sqkm: VH, covariance(independent)

mixed selfcompliance c.knowpeopleshare##i.VH i.female i.educ2 i.CtrlAge ib1.lived2 i.highPopulation i.burialtreat i.PGtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || n_sqkm: VH, covariance(independent)


margins VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))

**** MECHANISMS ****

mixed leadersanction i.VH i.female i.educ2 ib1.lived2 i.CtrlAge i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: VH, covariance(independent)

mixed leadersanction c.knowpeopleshare##i.VH i.female i.educ2 i.CtrlAge ib1.lived2 i.highPopulation i.burialtreat i.PGtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || n_sqkm: VH, covariance(independent)

margins VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))

mixed bandwagon i.VH i.female i.educ2 ib1.lived2 i.CtrlAge i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: VH, covariance(independent)

mixed bandwagon c.knowpeopleshare##i.VH i.female i.educ2 i.CtrlAge ib1.lived2 i.highPopulation i.burialtreat i.PGtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || n_sqkm: VH, covariance(independent)

margins VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))

mixed villagesanction i.VH i.female i.educ2 ib1.lived2 i.CtrlAge i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: VH, covariance(independent)

mixed villagesanction c.knowpeopleshare##i.VH i.female i.educ2 i.CtrlAge ib1.lived2 i.highPopulation i.burialtreat i.PGtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || n_sqkm: VH, covariance(independent)

margins VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))

*** Neighbor only: drop village head==1 ****
*drop if VH==1

mixed selfcompliance i.neighbor i.female i.educ2 ib1.lived2 i.CtrlAge i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor, covariance(independent)

mixed selfcompliance c.knowpeopleshare##i.neighbor i.female i.educ2 i.CtrlAge ib1.lived2 i.highPopulation i.burialtreat i.PGtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || n_sqkm: neighbor, covariance(independent)

margins neighbor, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))

*** MECHANISMS ***
mixed leadersanction i.neighbor i.female i.educ2 ib1.lived2 i.CtrlAge i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor, covariance(independent)

mixed leadersanction c.knowpeopleshare##i.neighbor i.female i.educ2 i.CtrlAge ib1.lived2 i.highPopulation i.burialtreat i.PGtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || n_sqkm: neighbor, covariance(independent)

margins neighbor, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))

mixed bandwagon i.neighbor i.female i.educ2 ib1.lived2 i.CtrlAge i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor, covariance(independent)

mixed bandwagon c.knowpeopleshare##i.neighbor i.female i.educ2 i.CtrlAge ib1.lived2 i.highPopulation i.burialtreat i.PGtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || n_sqkm: neighbor, covariance(independent)

margins neighbor, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))

mixed villagesanction i.neighbor i.female i.educ2 ib1.lived2 i.CtrlAge i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor, covariance(independent)

mixed villagesanction c.knowpeopleshare##i.neighbor i.female i.educ2 i.CtrlAge ib1.lived2 i.highPopulation i.burialtreat i.PGtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || n_sqkm: neighbor, covariance(independent)

margins neighbor, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))

********************************************************************************
******************************* APPENDIX ************************************
********************************************************************************

********* Main Models with Wealthy Sample (Table C.1) ******

* IMPORTANT: Rerun Code and Drop Poor Respondents from the Sample: drop if insuffinc == 1

mixed selfcompliance c.knowpeopleshare##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

mixed villagesanction  c.knowpeopleshare##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

mixed bandwagon c.knowpeopleshare##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

mixed leadersanction c.knowpeopleshare##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

************ OLS Regression Analysis with clustered standard errors (Table C.2) **********

reg selfcompliance c.knowpeopleshare##i.neighbor_VH i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief, vce(cluster numsqkm) 

reg villagesanction c.knowpeopleshare##i.neighbor_VH i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief, vce(cluster numsqkm) 

reg bandwagon c.knowpeopleshare##i.neighbor_VH i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief, vce(cluster numsqkm) 

reg leadersanction c.knowpeopleshare##i.neighbor_VH i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief, vce(cluster numsqkm) 

***** Table C.3-4: Analysis by local state vs. supra-local state actors and local customary vs. supra-local customary actors

* Local: neighbor / Local Non-State: Local chief/neighborhood Leader / Supra-Local Non-State: TA / Local State: Local Councilor / Supra-Local State: MP

* Compare Local vs. Supra-Local Non-State: drop state authorities and neighbors first
*drop if neighbor == 1
*drop if councilor == 1
*drop if MP == 1

mixed selfcompliance ib2.caul_q1 i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: caul_q1, covariance(independent)

* Compare Local vs. Supra-Local State: drop non-state actors and neighbors first
*drop if neighbor == 1
*drop if VH == 1
*drop if councilor == 1

mixed selfcompliance ib4.caul_q1 i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: caul_q1, covariance(independent)


********* Alternative thresholds for density of ties and community sanctioning, bandwagoning and leader sanctioning as DVs (Table C.5-7) **********

*80% threshold

mixed villagesanction  i.knowpeoplesqkm_binary##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

mixed bandwagon i.knowpeoplesqkm_binary##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

mixed leadersanction i.knowpeoplesqkm_binary##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

*60% threshold
mixed villagesanction  i.knowpeoplesqkm_binary2##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

mixed bandwagon i.knowpeoplesqkm_binary2##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

mixed leadersanction i.knowpeoplesqkm_binary2##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

*70% threshold
mixed villagesanction  i.knowpeoplesqkm_binary3##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

mixed bandwagon i.knowpeoplesqkm_binary3##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

mixed leadersanction i.knowpeoplesqkm_binary3##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)


*85% threshold
mixed villagesanction  i.knowpeoplesqkm_binary4##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

mixed bandwagon i.knowpeoplesqkm_binary4##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

mixed leadersanction i.knowpeoplesqkm_binary4##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

************************************************************************************
** Main models with compliance as DV and indivdiual social ties as control (reported in Appendix, Table C.8) **

mixed selfcompliance c.knowpeopleshare##i.neighbor_VH i.knowpeople i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

margins neighbor_VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, noci

mixed selfcompliance c.knowpeopleshare##i.neighbor_VH i.knowpeople i.female i.educ2 i.CtrlAge ib1.lived2 i.highPopulation i.burialtreat i.PGtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || n_sqkm: neighbor_VH, covariance(independent)

margins neighbor_VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, noci

********* Main Models with Interaction between ELF/ Poverty/ Rural/Urban and Leader Proximity (Table C.9) ******

mixed selfcompliance c.ELF##i.neighbor_VH i.burialtreat i.PGtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || n_sqkm: neighbor_VH, covariance(independent)

mixed selfcompliance i.highPopulation##i.neighbor_VH i.burialtreat i.PGtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || n_sqkm: neighbor_VH, covariance(independent)

mixed selfcompliance c.sqkinsuffinc##i.neighbor_VH i.burialtreat i.PGtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || n_sqkm: neighbor_VH, covariance(independent)

************ Fixed Effects with clustered standard errors (Table C.15) **********
encode sqkm, gen(numsqkm)
xtset numsqkm
xtreg selfcompliance i.neighbor_VH i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief, fe vce(cluster numsqkm) 

********* Alternative thresholds for density of ties and willingness to participate as DV (Table C.16) **********


*85% threshold

mixed selfcompliance i.knowpeoplesqkm_binary4##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

margins knowpeoplesqkm_binary4, dydx(neighbor_VH) asbalanced post

*80% threshold

mixed selfcompliance i.knowpeoplesqkm_binary##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

margins knowpeoplesqkm_binary, dydx(neighbor_VH) asbalanced post

* 70% thresholds

mixed selfcompliance i.knowpeoplesqkm_binary3##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

margins knowpeoplesqkm_binary3, dydx(neighbor_VH) asbalanced post

* 60% threshold

mixed selfcompliance i.knowpeoplesqkm_binary2##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

margins knowpeoplesqkm_binary2, dydx(neighbor_VH) asbalanced post

****************************************************************************************
********* Urban vs Rural Communities: Appendix Table C.17 and Figures C.6-7***********
****************************************************************************************

* Urban
*drop if highPopulation==0

*Rural
*drop if highPopulation==1

mixed selfcompliance i. neighbor_VH i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

*mixed selfcompliance i.neighbor_VH i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region c.knowpeopleshare || sqkm: neighbor_VH, covariance(independent)

mixed selfcompliance c.knowpeopleshare##i.neighbor_VH i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

margins neighbor_VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))

*********************************************************************************
********* Subgroup Analysis by Gender: Appendix Table C.18 and Figures C.8-9  *****
*********************************************************************************

* Female
*drop if female==0
*Male
*drop if female==1

mixed selfcompliance i. neighbor_VH i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

mixed selfcompliance c.knowpeopleshare##i.neighbor_VH i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

margins neighbor_VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))

******* Appendix C.19: Analysis with binary outcome *******
tab selfcompliance
gen self_binary = .
replace self_binary = 1 if selfcompliance == 3 | selfcompliance == 4
replace self_binary = 0 if selfcompliance == 1 | selfcompliance == 2

** Add level 1 variables 

mixed self_binary i.neighbor_VH i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

** Add level 1 and level 2 variables: not reported in main text **

*mixed self_binary i.neighbor_VH i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region c.knowpeopleshare || sqkm: neighbor_VH, covariance(independent)

* Cross-level interaction model **
mixed self_binary c.knowpeopleshare##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

margins neighbor_VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, recastci(rarea) ciopts(color(gs12))

** RECODING ***
*Do you think that people like you can have a say in what Your Traditional Authority/Tribal Chief/Chief does?
gen say_chief =.
replace say_chief = 1 if aurc_q1==2
replace say_chief = 0 if aurc_q1==1

gen say_VH =.
replace say_VH = 1 if aurc_q2==2
replace say_VH = 0 if aurc_q2==1

gen say_rel =.
replace say_rel = 1 if aurc_q3==2
replace say_rel = 0 if aurc_q3==1

gen say_MP =.
replace say_MP = 1 if aurc_q4==2
replace say_MP = 0 if aurc_q4==1

gen say_council =.
replace say_council = 1 if aurc_q5==2
replace say_council = 0 if aurc_q5==1

gen say_neighbor =.
replace say_neighbor = 1 if aurc_q6==2
replace say_neighbor = 0 if aurc_q6==1

*** Mulitlevel logistic regression results for Influence on leaders (C.20) *****

mixed say_chief i.female i.educ2 i.CtrlAge i.highPopulation c.knowpeopleshare i.n_LGPI_region|| sqkm:, covariance(independent)

mixed say_VH i.female i.educ2 i.CtrlAge i.n_LGPI_region i.highPopulation c.knowpeopleshare || sqkm:, covariance(independent)

mixed say_rel i.female i.educ2 i.CtrlAge i.n_LGPI_region i.highPopulation c.knowpeopleshare || sqkm:, covariance(independent)

mixed say_MP i.female i.educ2 i.CtrlAge i.n_LGPI_region i.highPopulation c.knowpeopleshare || sqkm:, covariance(independent)

mixed say_council i.female i.educ2 i.CtrlAge i.n_LGPI_region i.highPopulation c.knowpeopleshare || sqkm:, covariance(independent)

mixed say_neighbor i.female i.educ2 i.CtrlAge i.n_LGPI_region i.highPopulation c.knowpeopleshare || sqkm:, covariance(independent)

*********** Dropped from experment because no leader (Table D.4)***********
tab caul_q5
* drop no and do not know answers: 1,216 plus 685= 1901 (13.47%)

**** by leader type ****
gen dropped =.
replace dropped = 1 if caul_q5==1 | caul_q5==3
replace dropped = 0 if caul_q5==2

		  **  1 Village Head/Neighborhood Block Leader 
          **  2 Traditional Authority 
           ** 3 Local Councilor 
           ** 4 Member of Parliament 
          **  5 next door neighbor 

tab dropped if caul_q1 == 1
tab dropped if caul_q1 == 2
tab dropped if caul_q1 == 3
tab dropped if caul_q1 == 4
tab dropped if caul_q1 == 5
** we dropped village head=524 (18.82% of those who received the treatment), TA=403 (14.42), LC=453 (15.53), MP=359 (12.92), Neighbor=162 (5.69)

** by country 

tab dropped if caul_q1 == 1 & Zambia_c==1
tab dropped if caul_q1 == 2 & Zambia_c==1
tab dropped if caul_q1 == 3 & Zambia_c==1
tab dropped if caul_q1 == 4 & Zambia_c==1
tab dropped if caul_q1 == 5 & Zambia_c==1

tab dropped if caul_q1 == 1 & Malawi_c==1
tab dropped if caul_q1 == 2 & Malawi_c==1
tab dropped if caul_q1 == 3 & Malawi_c==1
tab dropped if caul_q1 == 4 & Malawi_c==1
tab dropped if caul_q1 == 5 & Malawi_c==1

tab dropped if caul_q1 == 1 & Kenya==1
tab dropped if caul_q1 == 2 & Kenya==1
tab dropped if caul_q1 == 3 & Kenya==1
tab dropped if caul_q1 == 4 & Kenya==1
tab dropped if caul_q1 == 5 & Kenya==1


************************************************************************************
**** Appendix E ****
************************************************************************************

******* Table E.1 Analysis with Additional Controls ************

mixed selfcompliance c.knowpeopleshare##i.neighbor_VH i.female i.educ2 i.CtrlAge ib1.lived2 i.highPopulation i.burialtreat i.PGtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || n_sqkm: neighbor_VH, covariance(independent)

margins neighbor_VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, noci

mixed leadersanction c.knowpeopleshare##i.neighbor_VH i.female i.educ2 i.CtrlAge ib1.lived2 i.highPopulation i.burialtreat i.PGtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || n_sqkm: neighbor_VH, covariance(independent)

margins neighbor_VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, noci

mixed villagesanction c.knowpeopleshare##i.neighbor_VH i.female i.educ2 i.CtrlAge ib1.lived2 i.highPopulation i.burialtreat i.PGtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || n_sqkm: neighbor_VH, covariance(independent)

margins neighbor_VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, noci

mixed bandwagon c.knowpeopleshare##i.neighbor_VH i.female i.educ2 i.CtrlAge ib1.lived2 i.highPopulation i.burialtreat i.PGtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || n_sqkm: neighbor_VH, covariance(independent)

margins neighbor_VH, at(knowpeopleshare=(0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1)) vsquish
marginsplot, noci


**** Table E.7 Social Ties Index - Preparation ****

* Important: calculate neighborhood ties index before dropping wealthy respondents afterwards

* vnin_q3: In this village/neighborhood would you say that you hardly know anyone etc.
* vnin_q4: When you think about your most immediate neighbors, are most, some, only a few or none related to you?
* vnin_q5: How often do you visit others in this village/neighborhood?
gen nties1 = vnin_q3 if vnin_q3 < 5

gen nties2 = .
replace nties2 = 1 if vnin_q4==4
replace nties2 = 2 if vnin_q4==3
replace nties2 = 3 if vnin_q4==2
replace nties2 = 4 if vnin_q4==1

gen nties3 = vnin_q5 if vnin_q5 < 5

* vnnp_q16: Are people from {name of respondent's village} more obligated to help each other, less obligated to help each other or neither more or less obligated to help each other than they are to help people from outside {name of respondent's village}?
* vnnp_q17: In general, are people from {name of respondent's village}  more worried, less worried or equally worried about being cheated when interacting with other people from {name of respondent's village} than they are when interacting with people from outside {name of respondent's village}?

gen nties4 = vnnp_q16 if vnnp_q16 < 4
gen nties5 = vnnp_q17 if vnnp_q17 < 4

* creating dummy variables *
gen dties1 = 0
replace dties1 = 1 if nties1 == 3 | nties1 == 4

gen dties2 = 0
replace dties2 = 1 if nties2 == 3 | nties2 == 4

gen dties3 = 0
replace dties3 = 1 if nties3 == 3 | nties3 == 4

gen dties4 = 0
replace dties4 = 1 if nties4 == 3

gen dties5 = 0
replace dties5 = 1 if nties5 == 3

**** Index: index of social ties with values between 0 and 5 (using dummies) ****
capture drop index1
gen byte index1 = (dties1 + dties2 + dties3 + dties4 + dties5)
hist index1, percent discrete

**** aggregated social ties (neighborhood level) ***
bysort n_sqkm : egen n_ties1=mean(index1)


***** ANALYSIS: Table E.7  

mixed selfcompliance c.index1##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

mixed selfcompliance c.n_ties1##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)


***** Main Random Slope Models after dropping the Kenya Sample (Table E.8)

*drop if n_LGPI_region == 4

** Add level 1 variables 

mixed selfcompliance i. neighbor_VH i.PGtreat i.burialtreat i.commonitortreat i.leadermonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)

* Cross-level interaction model **
mixed selfcompliance c.knowpeopleshare##i.neighbor_VH i.PGtreat i.burialtreat i.leadermonitortreat i.commonitortreat i.fewtreat i.alltreat i.noelder i.noVH i.nochief i.n_LGPI_region || sqkm: neighbor_VH, covariance(independent)



************************************************************
******* Correlations reported in Footnote ***************
************************************************************

*** Density with Income *****

pwcorr sqkinsuffinc knowpeopleshare
spearman sqkinsuffinc knowpeopleshare

**** Density with Social Inequality GINI ******


******* GINI: socioeconomic inequality - neighborhood level *******

* commands for Gini calculation *
*search descogini 
*search ineqdeco


***** using cover your needs *********

* in order to save the results from the ineqdeco command *
generate gini =.
qui ineqdeco insuffinc, by(n_sqkm)
foreach y in `r(levels)' {     
	replace gini = r(gini_`y') if n_sqkm==`y'	
}
tabdisp n_sqkm, c(gini)

pwcorr knowpeopleshare gini
spearman knowpeopleshare gini 


*********** Density with Ethnicity ***************
pwcorr ELF knowpeopleshare  
spearman ELF knowpeopleshare


*********** END ***********

